A combined K-means and hierarchical clustering method for improving the clustering efficiency of microarray

Among the microarray data analysis clustering methods, K-means and hierarchical clustering are researchers´ favorable tools today. However, each of these traditional clustering methods has its limitations. In this study, we introduce a new method, hierarchical K-means regulating divisive or agglomerative approach. The hierarchical K-means firstly employs K-means´ algorithm in each cluster to determine K cluster while operating and then employs it on hierarchical clustering technique to shorten merging clusters time while generating a tree-like dendrogram. We apply this method in two original microarray datasets. The result indicates divisive hierarchical K-means is superior to hierarchical clustering on cluster quality and is superior to K-means clustering on computational speed. Our conclusion is that divisive hierarchical K-means establishes a better clustering algorithm satisfying researchers´ demand.